Proxmox 6.0: USB devices disconnect and reconnect randomly

[Kadrim]

Hi there,

i just upgraded last night to Proxmox 6.0, since then i have random disconnects for my USB-attached drives. I currently have 3 HDDs connected via XHCI which worked flawlessly on Proxmox 5.x. After the upgrade i.e. the disc with the name /dev/sda gets disconnected after a short period of time (1-2 hours) and immediately reconnected but with a new name like /dev/sdf. This happens to all the drives attached via USB.

The dmesg outputs something like this:

[Tue Oct 29 08:20:52 2019] usb 2-2: USB disconnect, device number 2
[Tue Oct 29 08:20:52 2019] print_req_error: I/O error, dev sda, sector 0 flags 801
[Tue Oct 29 08:20:52 2019] sd 0:0:0:0: [sda] Synchronizing SCSI cache
[Tue Oct 29 08:20:52 2019] sd 0:0:0:0: [sda] Synchronize Cache(10) failed: Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK
[Tue Oct 29 08:20:52 2019] usb 2-2: new SuperSpeed Gen 1 USB device number 5 using xhci_hcd
[Tue Oct 29 08:20:52 2019] usb 2-2: New USB device found, idVendor=174c, idProduct=55aa, bcdDevice= 1.00
[Tue Oct 29 08:20:52 2019] usb 2-2: New USB device strings: Mfr=2, Product=3, SerialNumber=1
[Tue Oct 29 08:20:52 2019] usb 2-2: Product: USB 3.0 device
[Tue Oct 29 08:20:52 2019] usb 2-2: Manufacturer: Intenso
[Tue Oct 29 08:20:52 2019] usb 2-2: SerialNumber: 90900000000000003214
[Tue Oct 29 08:20:52 2019] scsi host5: uas
[Tue Oct 29 08:20:52 2019] scsi 5:0:0:0: Direct-Access Intenso USB 3.0 device 0 PQ: 0 ANSI: 6
[Tue Oct 29 08:20:52 2019] sd 5:0:0:0: Attached scsi generic sg0 type 0
[Tue Oct 29 08:20:52 2019] sd 5:0:0:0: [sdf] 1220942646 4096-byte logical blocks: (5.00 TB/4.55 TiB)
[Tue Oct 29 08:20:52 2019] sd 5:0:0:0: [sdf] Write Protect is off
[Tue Oct 29 08:20:52 2019] sd 5:0:0:0: [sdf] Mode Sense: 43 00 00 00
[Tue Oct 29 08:20:52 2019] sd 5:0:0:0: [sdf] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[Tue Oct 29 08:20:52 2019] sd 5:0:0:0: [sdf] Optimal transfer size 268431360 bytes
[Tue Oct 29 08:20:55 2019] EXT4-fs error (device sda1): ext4_find_entry:1455: inode #2: comm pvestatd: reading directory lblock 0
[Tue Oct 29 08:20:55 2019] EXT4-fs error (device sda1): ext4_find_entry:1455: inode #2: comm pvestatd: reading directory lblock 0
[Tue Oct 29 08:20:55 2019] EXT4-fs error (device sda1): ext4_find_entry:1455: inode #2: comm pvestatd: reading directory lblock 0
[Tue Oct 29 08:21:01 2019] sdf: sdf1
[Tue Oct 29 08:21:01 2019] sd 5:0:0:0: [sdf] Attached SCSI disk


I switched back to Proxmox 5.4, and now the system is running smoothly. I would suspect that something has changed in kernel 5.x.x that causes this issue.

Any ideas where to start?

 

[Chris]

Hi,
maybe some power saving configuration is causing the issue? https://www.kernel.org/doc/html/v4.14/driver-api/usb/power-management.html
Check the output of cat /sys/bus/usb/devices/<your device>/power/control

 

[Kadrim]

Hi Chris,

thanks for your reply

the output is just the same as with the previous version:

root@pve:/root# cat /sys/bus/usb/devices/2-2/power/control
on

so, autosuspend is not allowed.

The disconnect/reconnect also happens on a drive that can never be suspended because it has virtually no idle times (VM/CT storage).

My suspicion is, that something within kernel-space is causing this.

The module in question is nothing fancy:

00:15.0 USB controller: Intel Corporation Atom/Celeron/Pentium Processor N4200/N3350/E3900 Series USB xHCI (rev 0b)
Subsystem: Intel Corporation Celeron N3350/Pentium N4200/Atom E3900 Series USB xHCI
Kernel driver in use: xhci_hcd

 

[Kadrim]

just FYI: i updated to the latest testing kernel

Linux pve 5.3.7-1-pve #1 SMP PVE 5.3.7-1 (Wed, 23 Oct 2019 19:00:21 +0200) x86_64 GNU/Linux

but the problem still persists. switching to a 4.15.x kernel resolves the issue.

 

[Kadrim]

i also tried disabling autosuspend for all devices at boot time by adding

splash usbcore.autosuspend=-1

to the kernel command line. but no luck here.


Running Proxmox 6.0 with the "old" Linux pve 4.15.18-21-pve #1 SMP PVE 4.15.18-48 (Fri, 20 Sep 2019 11:28:30 +0200) x86_64 GNU/Linux Kernel works flawlessly.

I still cannot say what the change in kernel 5.X was, still comparing the kernel configs.

Anyone with another idea?

 

[Kadrim]

i switched my Hardware-Plattform from an AAEON Up-Squared (Intel N4200, 8GB RAM) to a Custom-Built ASRock Deskmini A300 (Ryzen 3400G, 32GB RAM) - using all the same USB-Devices.

With the new Server, this Problem does not happen. So it is not really solved but atleast bound to some form of Chipset-Problem present for the Intel N4200.

 

[germanb]

Hi @Kadrim i have similar issue with a ASMEDIA, its connect ok but then randomly disconnect. Do you solve this problem?

[ 5786.141579] usb 2-6: USB disconnect, device number 3
[ 5786.143217] sd 5:0:0:0: [sdf] Synchronizing SCSI cache
[ 5786.143238] sd 5:0:0:0: [sdf] Synchronize Cache(10) failed: Result: hostbyte=DID_NO_CONNECT driverbyte=DRIVER_OK
[ 5786.429687] usb 2-6: new SuperSpeed Gen 1 USB device number 4 using xhci_hcd
[ 5786.450519] usb 2-6: New USB device found, idVendor=174c, idProduct=55aa, bcdDevice= 1.00
[ 5786.450520] usb 2-6: New USB device strings: Mfr=2, Product=3, SerialNumber=1
[ 5786.450521] usb 2-6: Product: ASM1153
[ 5786.450522] usb 2-6: Manufacturer: asmedia
[ 5786.450522] usb 2-6: SerialNumber: 0123456789012
[ 5786.451522] usb 2-6: UAS is blacklisted for this device, using usb-storage instead
[ 5786.451523] usb-storage 2-6:1.0: USB Mass Storage device detected
[ 5786.451706] usb-storage 2-6:1.0: Quirks match for vid 174c pid 55aa: c00000
[ 5786.451800] scsi host5: usb-storage 2-6:1.0
[ 5787.454058] scsi 5:0:0:0: Direct-Access     asmedia  ASM1153          0    PQ: 0 ANSI: 6
[ 5787.454295] sd 5:0:0:0: Attached scsi generic sg5 type 0
[ 5787.454570] sd 5:0:0:0: [sdf] 1953525168 512-byte logical blocks: (1.00 TB/932 GiB)
[ 5787.454572] sd 5:0:0:0: [sdf] 4096-byte physical blocks
[ 5787.454955] sd 5:0:0:0: [sdf] Write Protect is off
[ 5787.454956] sd 5:0:0:0: [sdf] Mode Sense: 43 00 00 00
[ 5787.455246] sd 5:0:0:0: [sdf] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5789.688079]  sdf: sdf1
[ 5789.688922] sd 5:0:0:0: [sdf] Attached SCSI disk

Thanks

 

[Kadrim]

Sorry, but as i wrote earlier, i switched the platform altogether. so no, i think this was never fixed.

 

[Kadrim]

but you could try to run proxmox 5.x to see if the problem disappears.